Impact of vaccination on the association of COVID-19 with cardiovascular diseases: An OpenSAFELY cohort study – Nature.com

Characteristics of study cohorts

Among 18,210,937 people in the pre-vaccination cohort, 1,150,299 had a COVID-19 diagnosis during follow-up of whom 75,667 (6.6%) were hospitalised. There were 844,235 COVID-19 diagnoses (15,342 (1.8%) hospitalised) among 13,572,399 people in the vaccinated cohort and 162,103 (9,250 (5.7%) hospitalised) among 3,161,485 people in the unvaccinated cohort (Table1). Among 75,667 people in the pre-vaccination cohort who were hospitalised because of COVID-19 37,881 (50.1%) were hospitalised on the day of diagnosis and 6,278 (8.3%) were hospitalised the day after diagnosis: the remainder were hospitalised 228 days after diagnosis (TableS1). In the pre-vaccination cohort, the median age was 49 years (interquartile range (IQR) 3464), a slight majority (50.2%) were female, and 78.0%, 6.4% and 2.2% were recorded as being White, South Asian and Black ethnicities respectively. Differences between the vaccinated and unvaccinated cohorts reflected predictors of COVID-19 vaccine uptake20. The median (IQR) age was 54 (IQR 3968) years in the vaccinated cohort, compared with 36 (IQR 2847) years in the unvaccinated cohort. The proportions of females were 52.1% and 42.0% in the vaccinated and unvaccinated cohorts respectively, while the proportions recorded as of White ethnicity were 82.1% and 61.6% respectively, and the proportions living in the most deprived areas were 16.4% and 29.8% respectively. Compared with the vaccinated cohort, people in the unvaccinated cohort were more likely to be smokers, less likely to consult their GPs and less likely to have prior medical problems recorded (TableS2).

The numbersof events, person-years, and incidence rates per 100,000 person-years of vascular events before any COVID-19 diagnosis, after hospitalised COVID-19 and after non-hospitalised COVID-19, are presented, for each cohort and outcome, in Table2. There was a total of 212,557, 57,425 and 3,316 ATE in the pre-vaccination, vaccinated and unvaccinated cohorts respectively. The corresponding total numbers of VTE were 117,730, 29,107 and 3,178 respectively. In each cohort, the incidence of each arterial thrombotic and venous thrombotic event was higher after COVID-19 than before or without COVID-19. For each outcome and cohort, the highest incidence rates were after hospitalised COVID-19. Incidence rates were generally lower in the unvaccinated cohort than in the vaccinated cohort, as expected given that the median age of the unvaccinated cohort (36 years) was much lower than that of the vaccinated cohort (54 years).

Adjusted hazard ratios were estimated using Cox-proportional hazards models to quantify the associations between COVID-19 diagnosis (time-varying exposure) and a first cardiovascular event including arterial thrombotic, venous thrombotic and other cardiovascular events. Minimally adjusted models accounted for age, sex and region and maximally adjusted models accounted additionally for ethnicity, area deprivation, smoking status, number of GP-patient interactions and history of comorbidities. In each cohort, maximally adjusted HRs (aHRs) comparing the incidence of each outcome after COVID-19 diagnosis with the incidence before or without COVID-19 diagnosis were attenuated compared with age-, sex- and region-adjusted HRs (Tables35, TableS3). For all outcomes, hazard ratios were extremely high on the day of COVID-19 diagnosis (day 0), particularly among individuals hospitalised with COVID-19 on the day of diagnosis. The incidence of each outcome in each cohort was also elevated during weeks 14 after COVID-19 diagnosis, compared with before or without COVID-19 diagnosis. aHRs were lower in subsequent time periods than during weeks 14 after COVID-19 diagnosis, though they were generally greater than 1 throughout follow-up in each cohort (Figs.2 and 3, Tables35). aHRs during weeks 1-4 after COVID-19 diagnosis were substantially lower in the vaccinated cohort than in the pre-vaccination and unvaccinated cohorts, and generally remained lower than in other cohorts during weeks 428 (Figs.2 and 3, Tables35). For each outcome and in each cohort, aHRs were substantially higher after hospitalised than non-hospitalised COVID-19 (Fig.2, Tables3 and 4, TableS3).

Upper panels: Maximally adjustedhazard ratios and 95% CIsfor arterial thrombotic events. Lower panels: Maximally adjustedhazard ratios and 95% CIsfor venous thrombotic events. Left panels: all COVID-19 diagnoses: Middle panels: hospitalised COVID-19. Right panels: non-hospitalised COVID-19. The numbers of people in the pre-vaccination, vaccinated and unvaccinated cohorts were 18,210,937; 13,572,399 and 3,161,485 respectively. The numbers of COVID-19 diagnoses were 1,150,299 (75,667 hospitalised) in the pre-vaccination cohort, 844,235 (15,342 hospitalised) in the vaccinated cohort and 162,103 (9250 hospitalised) in the unvaccinated cohort. Maximally adjustedhazard ratios and 95% CIsare plotted at the median time of the outcome event within each follow up period in each cohort. Events on the day of COVID-19 diagnosis (day 0) were excluded. The numerical values of hazard ratios and their95% CIs are displayed in Tables3 and 4.

Upper left panel: Acute myocardial infarction. Upper right panel: Ischaemic stroke. Second row left panel: Pulmonary embolism. Second-row right panel: Deep vein thrombosis. Third row left panel: Heart failure. Third row right panel: Angina. Lower left panel: Transient ischaemic attack. Lower right panel: Subarachnoid haemorrhage and haemorrhagic stroke. The numbers of people in the pre-vaccination, vaccinated and unvaccinated cohorts were 18,210,937; 13,572,399 and 3,161,485, respectively. The numbers of COVID-19 diagnoses was 1,150,299 in the pre-vaccination cohort, 844,235 in the vaccinated cohort and 162,103 in the unvaccinated cohort. Maximally adjustedhazard ratios and 95% CIsare plotted at the median time of the outcome event within each follow-up period in each cohort. Events on the day of COVID-19 diagnosis (day 0) were excluded. The numerical values of hazard ratios and 95% CIsare displayed in Tables3, 4 and 5.

The incidence of ATE during weeks 1-4 after COVID-19 diagnosis, compared with before or without COVID-19 diagnosis, was elevated in the pre-vaccination and unvaccinated cohorts (aHRs 4.40 (95% CI 4.164.65) and 8.53 (7.2010.1) respectively) but less markedly elevated in the vaccinated cohort (2.09 (1.922.28)) (Fig.2, Table3). The incidence of ATE remained elevated during weeks 5-28 in the unvaccinated cohort (1.54 (1.162.04)) and up to weeks 53102 in the pre-vaccination cohort (1.22 (1.141.30)). During weeks 14 the aHRs for ATE were substantially lower in the vaccinated cohort than in the unvaccinated or pre-vaccination cohorts (ratios of aHRs 0.28 (0.250.32) and 0.36 (0.330.38) respectively, TableS4). Although attenuated, aHRs remained lower in the vaccinated cohort than in the unvaccinated or pre-vaccination cohorts during weeks 528 (ratios of aHRs 0.70 (0.520.94) and 0.73 (0.660.82), respectively).

The aHRs for ATE were substantially higher during weeks 14 after hospitalised COVID-19, versus before or without COVID-19 diagnosis (pre-vaccination cohort 12.1 (11.213.1), unvaccinated cohort 19.6 (15.624.5)) than after non-hospitalised COVID-19 (pre-vaccination cohort 2.70 (2.502.92), unvaccinated cohort 4.35 (3.365.64)). In sensitivity analyses restricted to primary diagnoses of ATE, aHRs during weeks 14 after hospitalised COVID-19 (including both day 0 and the rest of that period) were attenuated compared with aHRs for all ATEs (FigureS1). Estimated hazard ratios were similar in sensitivity analyses removing censoring at first vaccination in the unvaccinated cohort (TableS5). In additional analyses splitting follow-up during weeks 1-4 into shorter time intervals, hazard ratios for ATE declined steadily from days 16 to days 2127 after COVID-19 diagnosis, in all cohorts (TableS6).

The aHRs for VTE during weeks 1-4 after COVID-19 diagnosis, versus before or without COVID-19 diagnosis, were substantially higher than for ATE, particularly in the pre-vaccination and unvaccinated cohorts (aHRs 16.6 (95% CI 15.917.2) and 29.6 (26.732.9) respectively), but less markedly in the vaccinated cohort (4.87 (4.535.23)) (Fig.2, Table4). The incidence of VTE remained elevated, compared with before or without COVID-19 diagnosis, during weeks 528 in all cohorts and up to weeks 53102 in the pre-vaccination cohort (1.20 (1.091.32)). During weeks 14 the aHRs for VTE were substantially lower in the vaccinated cohort than in the unvaccinated or pre-vaccination cohorts (ratios of aHRs 0.17 (0.150.19) and 0.24 (0.230.26) respectively, TableS4). Although attenuated, aHRs remained lower in the vaccinated cohort than in the unvaccinated or pre-vaccination cohorts during weeks 528 (ratios of aHRs0 0.63 (0.490.80) and 0.61 (0.550.68), respectively).

The aHRs for VTE were substantially higher during weeks 14 after hospitalised COVID-19 (pre-vaccination cohort 88.5 (84.193.2), vaccinated cohort 46.7 (42.251.6), unvaccinated cohort 199.8 (174.7228.5)) than after non-hospitalised COVID-19 (pre-vaccination cohort 5.99 (5.616.40), vaccinated cohort 2.13 (1.912.38), unvaccinated cohort 7.64 (6.349.20)). The incidence of VTE was still markedly elevated during weeks 528 after hospitalised COVID-19 in the pre-vaccination, vaccinated and unvaccinated cohorts (aHRs 4.39 (4.014.80), 5.67 (4.636.94) and 7.50 (5.0711.1) respectively). In sensitivity analyses restricted to primary diagnosis of VTE, aHRs after COVID-19 diagnosis were attenuated compared with aHRs for all VTEs (FigureS2). This attenuation was particularly marked during weeks 14 (including both day 0 and the rest of that period) and after hospitalised COVID-19. In additional analyses splitting follow-up during weeks 14 into shorter time intervals, hazard ratios for VTE were generally similar during days 16 and days 713 after hospitalised COVID-19, then declined during days 1420 and days 2127 (TableS6). Hazard ratios after non-hospitalised COVID-19 did not markedly decline between days 1-6 and days 2127.

In each cohort, aHRs for acute MI during weeks 14 after COVID-19 diagnosis, versus before or without COVID-19 diagnosis, were similar to those for ischaemic stroke (Fig.3, Table3). In the pre-vaccination cohort, aHRs for acute MI remained elevated during weeks 2952 (1.16 (1.091.24)) and weeks 53102 (1.31 (1.191.45)), but the incidence of ischaemic stroke was only slightly elevated from 29 weeks onwards (aHR 1.16 (1.051.27) during weeks 53102). In all cohorts, aHRs during weeks 14 were markedly higher for PE (pre-vaccination 31.7 (30.333.1)), vaccinated cohort (9.10 (8.369.90), unvaccinated cohort 82.8 (72.794.3)) than for DVT, and aHRs for PE remained higher than for DVT during weeks 528 (Fig.3, Table4). By contrast, in the pre-vaccination cohort aHRs for DVT during weeks 29102 were higher than for PE.

The incidence of heart failure, angina, and subarachnoid haemorrhage and haemorrhagic stroke during weeks 14 after COVID-19 diagnosis was substantially elevated in each cohort, versus before or without COVID-19 diagnosis, although aHRs were lower in the vaccinated cohort than the pre-vaccination or unvaccinated cohorts (Fig.3, Table5). Compared with these outcomes, the incidence of transient ischaemic attack was less markedly elevated during weeks 14. Though greater than 1, aHRs for these four outcomes were markedly lower during weeks 528 than weeks 1-4 after COVID-19 diagnosis. In the pre-vaccination cohort, the incidence of heart failure during weeks 53102 was similar to the incidence before or without COVID-19 diagnosis (aHR 1.04 (0.981.11)). The incidence of angina and transient ischaemic attack was slightly elevated (aHRs between 1.10 and 1.16) and remained elevated during weeks 29102. aHRs for subarachnoid haemorrhage and haemorrhagic stroke were 1.32 (1.141.52) during weeks 2952 and 1.42 (1.151.76) during weeks 53102.

In subgroup analyses, aHRs for both ATE and VTE were generally lower in younger age groups, in females, and in those reporting white ethnicity (TablesS7, S8, FiguresS3, S4). Estimated excess risks of ATE 6 months post-COVID-19 diagnosis were 642, 229 and 718 per 100,000 people diagnosed with COVID-19 in the pre-vaccination, vaccinated and unvaccinated cohorts respectively (Fig.4, TableS9). Corresponding estimated excess risks of VTE were 797, 270, and 1094 per 100,000 people diagnosed with COVID-19, respectively.

Upper panels: Estimated absolute increase in risk for arterial thrombotic events. Lower panels: Estimated absolute increase in risk for venous thrombotic events. Left panels: pre-vaccination cohort. Middle panels: vaccinated cohort. Right panels: unvaccinated cohort. The numbers of people in the pre-vaccination, vaccinated and unvaccinated cohorts were 18,210,937; 13,572,399 and 3,161,485 respectively. The number of COVID-19 diagnoses was 1,150,299 in the pre-vaccination cohort, 844,235 in the vaccinated cohort and 162,103 in the unvaccinated cohort. Increases in risks were estimated within sex and age groups, and the estimated overall increase in risk is the average of these(shown in black), weighted according to the proportions in each sex and age group in the pre-vaccination cohort. Estimated excess events at 28 weeks are displayed in TableS9.

Originally posted here:

Impact of vaccination on the association of COVID-19 with cardiovascular diseases: An OpenSAFELY cohort study - Nature.com